|
Bo Gao
MSc. in AI Engineering @ Carnegie Mellon
University
Multi-modality · Image & Video Generation
· World Model
I received my Bachelor degree in the School of
Intelligence Engineering
at Sun Yat-sen University
(SYSU),
during which I also studied at The Chinese University of Hong Kong
(CUHK)
through a 2+2 program.
At SYSU, I was advised by
Prof. Xiaodan Liang.
I am now pursuing my Master degree at Carnegie Mellon University
(CMU), where I am advised by
Prof. Andrea Zanette.
My current research interests mainly lie in
Vision Language Model, image & video generation, and World Model.
📚 Passionate about Research
📷 Passionate about Photography
🎮 Passionate about Counter-Strike and Valorant
🎵 Passionate about Rap Music
🏀 Passionate about Basketball
📄 CV
💻 Github
🧠 Google Scholar
Email1: golbert2002 [at] 163 [dot] com
Email2: bogao [at] andrew [dot] edu
|
|
|
📢
News
-
📄 Submitted several papers about LLM to
ICLR 2027
(2026)
-
🏆 SLVR accepted as Poster at NeruIPS
2026
(2026)
-
🏆 One paper accepted as Findings at EMNLP
2026
(2026)
-
🏆 Bookagent accepted as Findings at ACL
2026
(2026)
-
🏆 Free-Mask accepted as Oral at ACM Multimedia
2025
(2025)
-
🎓 Received CMU Master offer and Dartmouth CS PhD offer for 2025
Fall
(2025)
|
|
🏅
Honors & Awards
- National Scholarship for 2021-2022
- First Prize Scholarship for 2021-2022
- Academic Innovation Scholarship for 2021-2022
- Special Research Scholarship for 2022-2023
- First Prize of National College Students Mathematical Competition 2022
- Meritorious Winner Prize of Mathematical Contest in Modeling (MCM) 2023
- Top 3% (24/929) in Kaggle: Image Matching Challenge 2024 – Hexathlon (CVPR'24 workshop)
[link]
🧩
Academic Service
- Reviewer: npj Digital Medicine (Nature Partner Journals), AAAI, NeruIPS, ICLR and CVPR.
|
|
Conference
|
|
SLVR: Structured Latent Visual Reasoning via Human-like Reasoning Flows
Albert Gao* , Bing Xue † , Andrea Zanette †
The Fortieth Annual Conference on Neural Information Processing Systems (NeruIPS), 2026
Poster
Accepted
paper
/
project
We propose the Linear Fitness Subspace (LFS) hypothesis: mutation-induced fitness variation encoded by PLMs is concentrated in a low-dimensional linear subspace of residue-level representation changes.
|
|
|
Linear Fitness Subspace in Protein Language Models Enables
Sample-Efficient Directed Evolution
SiYuan Ma, Canran Xiao, Zikai Xiao, Albert Gao , Liang He, Xuan-Yu Wang, Shuying Cao, Xiaojun Jia †
Empirical Methods in Natural Language Processing (EMNLP), 2026
Poster
Accepted
paper
/
project
We propose the Linear Fitness Subspace (LFS) hypothesis: mutation-induced fitness variation encoded by PLMs is concentrated in a low-dimensional linear subspace of residue-level representation changes.
|
|
|
BOOKAGENT: Orchestrating Safety-Aware Visual Narratives via Multi-Agent Cognitive Calibration
Bo Gao*, Chang Liu, Yuyang Miao, SiYuan Ma, Ser-Nam Lim †
The 64th Annual Meeting of the Association for Computational Linguistics (ACL), 2025
Poster
Accepted
paper
/
project
We explore collaborative AI agents with instruction tuning to generate long books with
improved realism and long-context consistency.
|
|
|
Free-Mask: A Novel Paradigm of Integration Between the Segmentation Diffusion Model
and
Image Editing to Improve Segmentation Ability
Bo Gao*, Jianhui Wang, Xinyuan Song, Yangfan He, Fangxu Xing, Tianyu Shi †
ACM Multimedia (ACM MM), 2025
Oral
Accepted
paper
/
code
We propose Free-Mask, combining diffusion-based segmentation with image editing to
create
realistic datasets that better match open-world settings while generating accurate masks.
|
|
|
Exploring Warping-Guided Features via Adaptive Latent Diffusion Model for Virtual
Try-on
Bo Gao*, Junchi Ren, Fei Shen †, Mengwan Wei, Zijun Huang
IEEE International Conference on Multimedia and Expo (ICME), 2024
Oral
Accepted
paper
/
code
We present ALDM, a warping-guided adaptive latent diffusion model for more
photorealistic virtual
try-on results.
|
|
|
Journal
|
|
Research on Two-Way Detection of YOLO V5s+Deep Sort Road Vehicles Based on the
Attention
Mechanism
Bo Gao*, Ronghui Zhang †
Journal of Physics: Conference Series, 2022
paper
We propose a "double vertical line" algorithm by introducing attention mechanisms into YOLOv5
and combining
them with Deep SORT-based tracking for robust vehicle detection.
|
|
|
🎓
Education
|
|
B.Eng. in Intelligence
Engineering @ Sun Yat-sen University
Sep. 2021 - Jun. 2025
GPA: 3.9 / 4.0, Ranking: 3 / 110
|
|
MSc. in AI
Engineering @ Carnegie Mellon University
Aug. 2025 - Now
|
|
|
